cross-posted from: https://libretechni.ca/post/559409
All words ending in “tion” or “ty” are both French and English. Apart from that, English gets many words from both Dutch and French that are similar. But there is no effort to exploit this because so many people are brainwashed to believe you should forget the existence of your 1st language when learning a new one.
I am firmly outside of that school of thought. When someone uttered the opening sentence of this post to me, I probably learnt ~6000+¹ words in French in 5 seconds. You cannot beat that. This would have taken years of playing charades using the popular immersion teaching style.
So the question is, are there any language learning tools whereby you specify two langauges and it produces a list or dictionary of true friends? The idea is that you can make a quick gain in vocabulary before progressing into unfamiliar/alienating words.
There are instances where I am writing a bilingual paper in English and French. The French column is a machine translation. Knowing some French (but not fluent), there are situations where the translation tool chooses a synonym for a true friend. If the machine had chosen a true friend, it would be easier for me to verify the quality of the translation and also easier for me to learn from. Considering my reader(s) are often native French and /possibly/ decent with English, there are also situations where I fail to choose an English word that would be easier for a francophone. So it would be useful as well if a translation tool would reverse the French back to English while trying to select true friends in English.
Furthermore, a reader of my French-English text may be a native Dutch speaker. So I would like an translation tool that adds some secondary gravity toward choosing English-Dutch friends when English-French falls short. Or another way to state this: I want a bilingual text that minimises the frequency of unique original words that are not borrowed by any of the relevant languages.
I realise gravitating toward true friends may cause a longer text in some cases, so I suppose I would also want to set a threshold of tolerance on additional words or syllables. In the end there would be some manual effort in the end anyway.
¹
grep -iE '(ty|tion)$' /usr/share/dict/american-english-huge | wc -l
No it’s not. Just take some language classes and take your own survey. It’s trivially verified.
It’s quite rare for a language class to use one language to learn another. Every single person I have surveyed believes (without evidence) that it’s better to learn a language without exploiting your mother tongue to learn a new language. Many language teachers are themselves instructed to avoid using the student’s mother tongue.
zaphod answered this well but I should add that 6k words are my count (from a dictionary), not the person who gave the tip. No one claimed that 6000 nouns results in “fluency”. (I scare-quoted fluency because B1 is where I’m at in French and I am nowhere near fluent; and I doubt B2 would get me there).
IIRC, “this guy” is Thomas Michael, a brit who produced audio tapes that teach French to English speakers. So there’s your source if you want to chase it up.
Does anyone else think Thomas Michael is full of shit?
Of course the AI bot would have to work that out and avoid such cases.